How Do Pain-Response Patterns In uence the Course of Acute Low Back Pain? A Longitudinal Cohort Study


 Background Pain-response patterns influence the daily activities and behavior of patients with low back pain (LBP). They are also a potential risk factor for pain persistence. Previous studies identified three subgroups of pain-response patterns: avoidance, endurance, and adaptive. In this study, we investigated the interaction effects of distinct pain-response patterns on pain intensity, disability and movement control impairment (MCI). Methods The 66 participants in this prospective cohort study had acute LBP, assessed at four time points over 6 months. Demographic, clinical and psychosocial data were collected, and MCI was examined.Results The results showed no significant between-group interactions of the different pain-response patterns with pain intensity, disability or MCI. However, within-group decreases in pain intensity and the disability level from the acute to the subacute phase were identified. In addition, the relative risk of persistent LBP was higher in the endurance response subgroup than in the adaptive subgroup at 6 months. MCI remained stable and at a low level throughout the observation period and did not correlate with pain or disability.Conclusions Screening for pain-response patterns may lead to a better understanding of patients’ underlying motivation in the broader context of their valued goals. Our findings provide further evidence of altered movement control throughout the course of LBP and thus the importance of early interventions. However, MCI tests allowing more subtle assessments of movement control are needed. Finally, patients with an acute episode of LBP may benefit from a multidimensional clinical approach that takes into account both physical and psychological health.


Introduction
Acute low back pain (LBP) has a favorable prognosis, with the majority of patients recovering within 6 weeks [1]. However, some patients will develop disabling persistent or recurrent LBP [e.g. 1,2]. In the absence of an identi able speci c pathology, the nature of the underlying multidimensional pain drivers has raised both clinical and research interest [3][4][5]. Several studies have shown that the interactions of psychosocial and biomechanical factors are important in the development of LBP [5][6][7][8]. Although LBP remains a highly individual experience, the pain responses are of two basic types, provocative or protective [9][10][11][12], and derive from psychological aspects that include cognition (e.g., beliefs, thought suppression), affect (e.g., fear/anxiety, mood), and behavior (e.g., persistence, avoidance, pacing) [11][12][13]. Based on these factors, three subgroups of pain-response patterns have been de ned: fear-avoidance responses (FAR), endurance responses (ER), also described as overactivity [11,14], and adaptive responses (AR) [15]. The FAR pattern consists of activity avoidance due to the conviction that certain activities will worsen the pain. However, the resulting structural/functional deconditioning and, frequently, the associated depression may promote pain persistence. In the ER pattern, overactivity and task persistence despite pain may lead to a dramatic increase of pain intensity that prohibits further activity Page 3/18 [6,15]. The long-term consequences of this strategy include pain exacerbation due to overuse or softtissue micro-trauma [6,12]. The AR pattern involves coping strategies aimed at achieving an optimal balance between activity and regeneration and often results in favorable outcomes [6,12].
In contrast to the AR pattern, the FAR and ER patterns pose higher risks of persistent pain and disability, as the maladaptive changes in musculoskeletal structures impede accurate movement control, albeit through opposite triggers [9,11,12]. The resulting changes in lumbo-pelvic kinematics may lead to movement control impairment (MCI) [16][17][18], an underlying driving mechanism that develops secondary to pain [5,16,19,20]. MCI is characterized by the inability to actively control the position of the low back during functional movements [18], either because active movement of the spine is limited by increased muscle activities that place more strain on the spine [16,21] or because of an impaired ability to actively control movement of the back in a neutral position during functional activities [18]. Although interventions aimed at improving MCI may be effective in reducing pain and disability, the factors that lead to the initial development of MCI in the course of LBP are largely unknown. Furthermore, the contribution of MCI itself to persistent pain merits further investigation [4,5,8,[22][23][24]. In this study we explored the time-related effects and interactions of pain-response patterns on pain intensity, disability and MCI in the course of acute LBP.

Study Design
In this prospective cohort study patients were assessed at 4 time points: ≤ 4 (baseline), 8, 12, and 26 weeks after the onset of acute LBP. At each time point, clinical tests were performed. Patient-reported outcomes were collected using an online survey that queried medical history, demographic variables, LBP-associated pain and the disability level. All participants provided informed consent prior to the start of the examination. The Medical Ethics Committee of the Canton of Zurich, Switzerland (BASEC-no. 2016-02096) approved the study protocol. All experiments were performed in accordance with relevant guidelines and regulations

Participants
The 66 patients with acute LBP were recruited from private physiotherapy practices, two hospital outpatient practices, and the University Campus using advertisements and mailing lists between December 2017 and October 2018. Inclusion criteria were 1) onset of acute LBP with or without radiation of pain within the last 4 weeks, 2) in case of recurrent LBP: pain-free episode at least 3 months prior to pain onset, 3) age 18-65 years and 4) su cient German language skills to ll out the questionnaires. The exclusion criteria were: 1) signs of serious pathology, 2) major psychiatric illness necessitating psychiatric/psychological consultations, and 3) current pregnancy or recent birth (past 12 months). Of the 89 candidates screened for eligibility, 66 were enrolled at baseline. Of this group, 86% (n= 57) were available at week 8, 83% (n=55) at week 12, and 79% (n=52) at week 26. The most common reasons for loss-to-follow up were insu cient adherence to the schedule of clinical tests or questionnaire completion and pregnancy (see Figure 1 Supplementary Material). There were no differences between participants and dropouts at baseline.
Patient-reported outcome measures. The online survey was sent as a link by email and the patients were asked to complete the questionnaires within 2 days. Patients who did not complete the questionnaires within this time were sent an electronic reminder, followed by a phone call in case of no response.
Demographic and clinical variables. Baseline sociodemographic data on sex, age, marital status, education level, and number of previous LBP episodes were collected, as well as information on the use of medical services, and sick leave.
Pain intensity. Present pain, worst pain and average pain intensity of the past week were measured at all four time points using a numeric self-rating scale (NRS) ranging from 0 to 10. The mean pain intensity was then calculated. Areas of referred pain were also documented.
LBP associated disability. The Oswestry Disability Index (ODI) quanti es overall functional disability. The German version of the ODI has excellent test-retest reliability (r = 0.96) and strong correlations with the Roland Morris Disability Questionnaire (r = 0.8) in patients with chronic pain [25].
Depression. The 21-item Depression Anxiety Stress Scale (DASS) questionnaire was designed to assess the severity of depression, anxiety and stress using three subscales [26]. Depression has a cutoff of >10 sum score on the subscale depression, with high internal consistency (Cronbach's alpha 0.91) and construct validity (r = 0.68) with the Beck Depression Inventory [27].
Pain-response patterns. The Avoidance-Endurance Questionnaire (AEQ) was designed to assess painresponse patterns according to four different subgroups: adaptive response (AR), fear-avoidance responses (FAR), distress endurance responses (DER), and eustress-endurance responses (ER) based on the Avoidance-Endurance model [15]. The subgroups were established using the thought suppression scale (TSS) and the behavioral endurance scale (BES), together with the depression subscale of the DASS [28] (S1 Table).
Clinical examination. At each time point, clinical data on pain, MCI, and neurological status (sensory neuron and motor responses) were collected to screen for serious pathology and re-control the exclusion criteria. All investigators underwent a 2-hour training session to ensure the intra-and inter-tester reliability of the clinical tests and study procedure.
Lumbar movement control abilities were evaluated using a battery comprising six active direction-speci c movement tests for exion, extension and rotation. Investigators provided verbal movement instructions and repeated them once in case of incorrect movement execution. Investigators then rated the execution as correct (negative = 0 points) or incorrect (positive = 1 point), with total scores ranging from 0 to 6 and a high score interpreted as high impairment of movement control [29]. The discriminative validity of the test battery in patients with LBP vs. healthy controls was previously demonstrated, with mean scores of 2.2 in patients with LBP and 0.75 in controls (p<0.05) and good to excellent intra-and inter-tester reliability [18,29,30].

Statistical analyses
Descriptive statistics were used to screen for disproportional participant characteristics and data outliers. A 2-way repeated measures analysis of variance (ANOVA) was then conducted with pain-response patterns as the between-subjects factor, time as the within-subjects factor (baseline, 8, 12 and 26 weeks) and pain intensity, disability and MCI as outcome variables. In case of signi cant time effects, Bonferroni post-hoc tests were applied to determine within-subject differences. Mean values were substituted for missing values. Because age and sex were considered to impact pain, disability and MCI, adjustments were made for their effects using a repeated measure analysis of covariance (ANCOVA). A further adjustment was included for the number of previous LBP episodes (0, 1-2, 3-4, and >4). The relative risk (RR) of pain persistence in each subgroup was examined at 12 and 26 weeks based on two pain categories: < NRS 3/10 and ≥NRS 3/10. Finally, the bivariate correlations of pain intensity, disability, and MCI were calculated using Spearman's correlations. The signi cance level was set at p < 0.05. SPSS 25.0 (Statistical Package for the Social Sciences) was used for all statistical analyses.

Sample characteristics
The 66 patients enrolled at baseline had a mean age of 39.7 (SD 12.8) years. The majority were female (53%). Education levels widely differed, with the majority holding a university degree (60.6%). The average pain intensity at baseline was 4.3 (SD 1.8). The mean reported days of work absence due to sickness was 1.8 (range 0-14). Most participants reported at least one previous episode of LBP. Depression, anxiety and stress scores were low. Based on the AEQ classi cation of pain-response patterns, the ER subgroup comprised 38 (57.6%) patients, the AR subgroup 24 (36.4%) patients, and the DER subgroup 1 patient. The mean MCI score at baseline indicated impairment. Table 1 shows the sample characteristics at baseline.    Figure 2 Supplementary Material).

Discussion
This study investigated the time-related effects and interactions of pain-response patterns on pain intensity, disability and MCI after an episode of acute LBP, based on simultaneously collected clinical data, self-reported outcomes and trunk movement control determinations (MCI score). Pain intensity and disability decreased signi cantly over time, while MCI scores were consistently low. Contrary to our expectations, neither ER nor AR had an impact on the outcome variables, although the risk of persistent pain was higher in the ER than in the AR subgroup.
Most acute LBP episodes are transient [1] with modest or no aftereffects [3]. In our patients, pain intensity and disability decreased signi cantly within the rst 8 weeks but remained stable thereafter, until the last follow-up at 26 weeks. Our results are consistent with previous descriptions of LBP trajectories, including reports on the natural course of LBP [31][32][33][34]. However, time-averaged reporting for pain and disability does not re ect individual variability [35], as the distinct trajectories of LBP may include a less pronounced pain decrease and the persistence of pain at an overall high level in patients with severe or uctuating pain [31][32][33][34]. Despite the different pain characteristics reported by our patients, the LBP episode assessed in our study was assumed to be non-transient, and while the pain intensity remained at a low level, it may well have in uenced the quality of life of our patients. In our study, previous episodes of LBP had no in uence on pain intensity and disability, but together with awkward posture and longer sitting time they may be predictive of recurrent LBP within 12 months [36].
During a LBP episode, pain responses de ne patterns of daily activities and behavior. As such, they are a potential risk factor for pain persistence. The pain-response patterns de ned by the AR, ER and DER subgroups were constructed using AEQ subscales and a depression inventory. The presence of depressive symptoms requires classi cation into the DER or FAR subgroup. The only patient in our study with depression was categorized in the DER subgroup; none of the patients exhibited the FAR painresponse pattern, recognized as a signi cant prognostic factor of persistent LBP [37,38]. However, the reported prevalence of the FAR pattern is < 10% [12], whereas that of ER is ~40% [12,13]. The absence of the FAR pattern in our patients re ects the fact that major psychiatric illness requiring treatment was among the exclusion criteria. Rather, the rate of self-reported depression among our patients was low, as previously shown for patients in the general population with acute LBP necessitating primary care [32,39]. Indeed, when followed over one year, patients with low pain intensity and good mental health seem to experience an improvement of LBP, whereas the opposite features are associated with persistent highlevel pain [33]. However, the validity of these causality relationships and their prognostic implications require further investigation [31].
The majority of our patients exhibited the ER pain-response pattern. Recent research demonstrated the harm caused by ER behavior with respect to pain intensity and the long-term consequences [7,9,11,12].
Conversely, bene cial effects related to optimal pain coping strategies, such as avoidance adapted to the situation and low persistence behavior, have been reported for the AR pattern [12,40]. Nonetheless, the outcomes differed from those expected based on the features of these two subgroups, as ER and AR patients did not signi cantly differ regarding pain intensity and disability, although both scores were higher in the ER subgroup. Nonetheless, a calculation of the RR for ER compared to AR patients revealed an elevated risk of pain persistence in the former at 12 weeks and especially at 26 weeks. This is consistent with the in uence of ER-type pain responses on pain intensity and the negative long-term consequences thereof [7,9,11,12], as these patients may feel con icted by the desire for pain relief but also to maintain their usual activities [7]; see personal goals [10]. In fact, the meta-analysis of Andrews et al. (2012) linked endurance responses with elevated pain and poor functioning in daily tasks [9].
The majority of our patients presented with MCI, with a mean score that remained unchanged throughout the 26-week follow-up. Similar results were reported in a study of patients with acute, subacute, and chronic LBP [29]. However, an additional analysis by those authors in a case control study revealed signi cant differences in MCI, with scores increasing successively from acute to chronic LBP [18]. The longer observational time in our study may have masked differences in the scores of our patients. We found that MCI does not differ in patients with AR and ER responses, and therefore that these responses do not directly contribute to movement control. However, this nding is di cult to explain, given that the overuse that characterizes ER-type responses results in maladaptive changes in musculoskeletal structures that impede accurate movement control. We therefore hypothesize that a time-dependent persistence in pain intensity in the ER subgroup limits movement control but this behavior is not re ected in the MCI tests, especially since pain intensity was not associated with the MCI scores. In fact, evidence for the construct validity of the MCI tests is lacking. Most MCI assessments are based on interpretations of the quality of movement execution and the identi cation of detrimental mechanism involving the lumbar spine [30]. Indeed, prospective studies revealed a reduction in disability scores in patients treated with individual movement control exercises [(20) 36]. Whether pain reduction was achieved in those patients could not be ascertained from the data. Further studies are needed to determine whether pain reduction is achieved by motor control interventions, as in the absence of prospective studies the relationship between MCI and pain remains unclear [41][42][43][44]. The adaptive changes in trunk motor control are also poorly understood, given the variation among individuals depending on tissue integrity and task performance [7,22,39]. Despite the gradual shift from categorized movement control intervention toward a more multidimensional understanding of the complexity of LBP [45], further work in this area is mandatory to link MCI to pain provocation [45][46][47], as movement control may turn out to be of less relevance over the course of acute LBP.

Conclusion
In conclusion, LBP is multidimensional in terms of its mechanisms and manifestations, both of which must be recognized in its treatment. Our investigation of the psychological and physical aspects of LBP revealed a need for a comprehensive clinical approach to LBP and a continued exploration of the provocative underlying factors [47,48]. A key strength of our study was its cohort design and the simultaneous assessments of clinical data, self-reported outcomes and trunk movement control that were performed at each of the four time points. However, due to the small sample size, caution must be applied in interpreting our ndings, which may not be representative. Notably, our population had a high educational level whereas people with a lower educational level or in poor general condition due to acute pain may be less compliant with the demands of study participation [49]. In addition, as our patients represented only the AR and ER pain-response patterns, studies of FAR and DER responses and their possible effects on outcome variables are still needed. Nonetheless, our ndings indicate that screening pain-response patterns may provide a better understanding of patients' underlying motivation in managing their LBP and thus an improved therapeutic strategy [12,47]. For example, AR patients at lower risk of persistent LBP may need less treatment, the bene ts of which would include cost savings [12,50,51]. A recognition of these patients requires an in-depth understanding of the psychological aspects of LBP and adequate communication skills in clinical practice [52][53][54][55]. Given the altered movement control throughout the course of LBP, also identi ed in this study, early interventions of motor control training should be considered [45,56,57]. Furthermore, our results provide additional evidence that MCI tests do not allow for more subtle assessments of movement control. Moreover, they also highlight the importance of a therapeutic shift toward a multidimensional approach to acute episodes of LBP, one that takes into account the physical and psychological characteristics of these patients. Availability of data and materials: Open access will be granted on Zanodo for the datasets used and/or analyzed during the current study or from the corresponding author on reasonable request.
Competing interests: The authors declare that they have no competing interests.
Funding: This work was supported by the Swiss National Science Foundation (No. 320030_173297 /1).